The Truth About Soy – Supplements and Nutrition – Forums

T Nation has been reporting on the hormone-disrupting effects of soy since the year 2000, but some still say it’s a myth. It’s not. Here’s why.

Some say it’s a myth that soy can lower testosterone levels and have estrogenic effects in men. Well, it’s not a myth. Sure, the effects of soy require qualification and context, but they’re real.

Randomized controlled trials are the strongest type of study. They provide an intervention, randomize subjects into an active versus control group, and blind the participants and researchers as to who’s receiving an active or control. This helps us assess whether the independent variable (soy isoflavones) can impact the dependent variable (testosterone).

Here are those studies in support of the soy/testosterone relationship:

In a randomized controlled trial with crossover, 35 healthy men consumed a protein supplement consisting of either milk protein isolate (MPI), low isoflavone soy protein isolate (LISPI), or high isoflavone soy protein isolate (HISPI) for 57 days, followed by a 28-day washout before switching proteins (1). The MPI, LISPI, and HISPI groups consumed 0, 1.64, and 61.7 mg/day of soy isoflavones, respectively.

The LISPI and HISPI groups experienced a decreased serum dihydrotestosterone (a 5 alpha-reduced testosterone metabolite responsible for androgenic effects in certain tissues) and DHT/testosterone ratio as compared with the MPI group at day 57. Serum testosterone was decreased only in the LISPI group as compared to the MPI at day 29 and HISPI group and only at day 57.

Interestingly, the study also looked at whether there were any differences in these values between equol producers and those who weren’t. Equol is a metabolite of one of the two main soy isoflavones – genistein – that’s more estrogenic than daidzein. It reported no differences (2, 3).

In a randomized trial with crossover, 42 healthy men, ages 35-62 years, consumed either 150 grams of lean red meat per day for 4 weeks or 290 grams of tofu (soy isoflavone content not reported but probably around 66 mg/day) (5).

While total testosterone levels weren’t different between the two groups, sex hormone binding globulin (SHBG) – a protein that binds to testosterone, making it unavailable – was 3% higher in the tofu group, while the free androgen index (FAI) was 7% lower.

In another trial, 19 healthy men consumed scones made with either soy flour (containing 120 mg/day of soy isoflavones) or wheat flour for 6 weeks, followed by a 6-week washout period (6). The soy flour eaters experienced a 5.7% decrease in total testosterone.

In still another trial, 58 men, ages 50-85 years, at high risk for developing prostate cancer, were assigned to one of three groups of protein supplementation (7). One group received 40 grams of soy protein isolate containing approximately 107 mg/day of isoflavones. The next group received soy protein isolate with greatly reduced isoflavone levels (< 6 mg/day). A third group got milk protein isolate (no soy isoflavones).

After 6 months of supplementation, androgen receptor (AR) expression in the prostate was lower in the soy protein isolate with isoflavones, while AR expression was increased in the milk protein isolate group. The decrease in AR expression is a positive effect regarding prostate cancer.

One more: Healthy younger males aged 21 years were given 20 grams per day of whey protein, soy protein isolate (26.3 mg of soy isoflavones), or a placebo each day for 14 days (9).

The subjects then performed resistance exercise and had post-exercise hormone levels evaluated. From 5 to 30 minutes post-exercise, the soy eaters had significantly lower testosterone levels than those consuming whey protein and placebo. This suggests that soy protein isolate may blunt the normal increase in testosterone experienced after lifting (8).

Open-label studies (where the researchers and subjects are aware of what they’re given) without a control group and randomization are prone to confounders. More caution is warranted, but they can still provide potentially useful info.

In one such study, 12 healthy men ages 25-47 years were given 56 grams per day of soy protein isolate for 28 days (10). Total isoflavone content wasn’t evaluated but may have been around 51 mg (11).

After 28 days, total testosterone decreased by an average of 19% and began to increase again within two weeks after discontinuing the supplement. Luteinizing hormone (LH), which provides the initial signal for testicular testosterone production, also decreased, albeit not to a statistically significant extent.

In another study, 28 healthy men ages 30-59 years were given 60 mg/day of a soy isoflavone supplement for 3 months (12). While estradiol didn’t change when comparing baseline values to those taken after 3 months, SHBG increased while DHT and free testosterone decreased, but only in the group of men classified as equol producers.

Case reports are good for identifying potential hypotheses, but because they’re subject to so many uncontrolled variables, they generally don’t allow scientists to conclude cause-effect.

There’s one case report on a 19-year-old male who, aside from having type I diabetes mellitus, was otherwise healthy. Physicians reported he was suffering from erectile dysfunction, lack of libido, and suppressed total and free testosterone (13).

It turns out he was consuming very high amounts of soy products, with an estimated daily total soy isoflavone intake of 360 mg for a year. Once he quit the soy, his symptoms normalized over time.

The physicians attribute the effects experienced by this young man to soy intake. This seems reasonable, but he was also consuming atorvastatin (a statin drug), which may decrease testosterone (14-16).

The authors didn’t report whether the patient also ceased consuming his medications, including atorvastatin. Knowing that would’ve helped to solidify the case against soy. Nonetheless, it’s an interesting case.

There are also a couple of case reports of soy-associated gynecomastia with excessive consumption (17,18). In one case, the total soy isoflavone was around 360 mg/day (17).

This case was also unusual due to the substantial rise of the estrogens, estradiol and estrone. Studies generally haven’t found an increase in estradiol or estrone with soy consumption, although some have, oddly enough, with soy protein supplementation that appears to be unrelated to soy isoflavone content (1,7,19).

If related to isoflavone content alone, a potential explanation might be increased aromatase activity after high concentrations of genistein are reached in certain tissues (at least 100 nmol), leading to the formation of excess estrogens (20).

However, this wouldn’t explain how soy protein with minimal isoflavone levels could lead to increased estrogens. It may simply be a chance finding or this could represent yet another unexplained discrepancy, but it’s most likely the former. In any event, gynecomastia is likely to be very rare and probably involves other variables.

Finally, a more recent case report featured a 54-year-old male that developed secondary hypogonadism after drinking 1.2 L per day of soy milk containing an estimated 310 mg of isoflavones for a period of 3 years (110).

His LH, total, and free testosterone levels were dramatically suppressed but normalized after he stopped drinking it. He also suffered from erectile dysfunction and gynecomastia, which improved after he ditched the soy milk.

However, his estradiol levels weren’t elevated and were below the normal range, presumably due to the lack of testosterone production. He was following a low-carbohydrate diet, but there was no report of any other abnormal medical history or medications that might explain his symptoms.

In an observational study of Japanese men, an inverse association was found between total and free testosterone and soy product intake (21). However, these types of studies are also prone to potential confounders where variables other than the one you’re evaluating are involved.

A recent study found a significant association between urinary levels of soy isoflavones (especially genistein) and decreased plasma testosterone levels (111).

Admittedly, there are several randomized controlled trials in men that haven’t displayed an effect on testosterone levels (24-29). Intake levels of total soy isoflavones have ranged from 24-138 mg/day, with typically around 65 mg/day. Total genistein has ranged from 36-54 mg/day.

In a prospective observational study, a higher urinary excretion of soy isoflavones was associated with decreased semen quality (39), while another study only found a decrease in associated sperm concentration (40).

An observational study in Chinese men also revealed similar findings concerning sperm quality, which is consistent with data in Japanese men (41,42). However, a recent cross-sectional observational study failed to find any such association (111).

It’s important to note, though, that randomized, controlled trials have failed to find an adverse effect of soy isoflavone supplementation on semen (43,44). One study also evaluated those that were equol producers and found no association with semen quality (43).

Why do some studies find testosterone drops while others don’t? There are several possible explanations, many of which include differences in methodology: strength of study design, soy isoflavone intake level and formulation, bioavailability of the isoflavones in the food or supplement, study subject characteristics, testosterone assay methods, etc.

However, while the evidence isn’t conclusive, there are data to support the hypothesis that equol production is one possible determinant of the estrogenic effects of soy isoflavones (45). Remember, equol is as potent or more potent as the major soy isoflavones, genistein and daidzein, respectively.

Soy isoflavones can exert direct estrogenic effects in the form of ER-beta agonism (a compound’s ability to bind to and activate the estrogen receptor subtype beta). This agonism from soy isoflavones is achievable with lower doses of soy isoflavones (genistein in particular), around 0.25 to 1 mg/kg, although it could be higher if factoring in potential protein-binding.

This ER agonsim isn’t necessarily a bad thing. It’s linked with beneficial effects on cardiovascular, bone, and prostate health. Although, the potential benefits of soy may also be mediated through mechanisms unrelated to binding to and activating ER-beta.

Unlike the direct estrogenic effects, hormonal effects (decreased testosterone) probably require doses much higher than 1 mg/kg before small effects are seen. Of course, this is complicated by both the large inter-individual variation in how the soy isoflavones are absorbed and metabolized, as well as potential differences between individuals in terms of tissue sensitivity to the soy isoflavones.

In any event, doses of soy isoflavones of around 1 mg/kg (and between 0.25 to .75 mg/kg genistein) can produce average unconjugated genistein peak plasma concentration of around 10-20 nmol. These concentrations are effectively maintained with daily consumption. (34,35,66-69).

These concentrations are well within the range where genistein can act as an ER-beta agonist in vitro (3,31,33,70-72). Furthermore, plasma concentrations alone can be significantly lower than the tissue concentrations, depending on how the soy isoflavones are distributed throughout the body and how much is bound.

The data for exactly how much we should expect to be bound aren’t clear. Many studies use concentrations well beyond what’s relevant to administration in humans. But based on in vitro data showing a higher affinity at lower concentrations and animal model data, it’s safe to assume that at least 50% is protein-bound at any given time (82,104-108). So, even higher doses of soy isoflavones and genistein specifically may be needed before ER-beta agonism is likely.

While the concentrations required for direct estrogenic activity are attainable with a reasonable level of soy intake, the dose required for effects on testosterone is likely higher, probably at least 1.5 to 2 mg/kg, before small effects are seen in sensitive individuals.

However, even greater amounts may be required when factoring in protein-binding. In vitro data evaluating genistein show the concentrations necessary for inhibiting or interfering with testosterone synthesis are in the range of around 40-90 nmol (76,77).

This would also indicate that very high doses, presumably between 4-8 mg/kg, would be required before a substantial decrease in testosterone would be seen (78-80). This fits with the doses consumed in case reports. While other mechanisms have been proposed to account for the potential lowering of testosterone, those often require very high concentrations (1 umol or much higher) of the unconjugated isoflavone, which isn’t attainable even with extremely high consumption.

Finally, aside from enzymatic mechanisms likely only affected with high doses, the estrogen receptor alpha (ER-alpha) subtype could also induce negative feedback if activated by isoflavones (81). However, genistein, for example, would require much higher doses to activate the ER-alpha versus the ER-beta.

In particular, doses between 3-8 mg/kg would potentially be needed before plasma concentrations could reach concentrations capable of making this a potential issue as far as lowering testosterone via reduced gonadotropin secretion (3,31,33,70-72,79,80,82).

Although, these concentrations in plasma would also need to be attained in the brain – the hypothalamus and pituitary – which isn’t necessarily the case for the parts of the brain that are of concern for hormone signaling which exist outside the blood-brain barrier.

And remember, when factoring in protein-binding, the 3-8 mg/kg dose may be underestimating the amount required for concentrations necessary for ER-alpha agonism (82,104-108).

A recent in vitro assay purporting to reveal “estradiol equivalents” or E2Eq for various food items indicated that a cooked tofu burger (198 g) could potentially contribute E2Eqs of 19 pg/mL (the normal range for a man listed was 10-40 pg/mL) (83).

The authors also reported that the total amount of estrogenic activity for the tofu burger was three times that of a 125 mg dose of the estrogen replacement drug Premarin, which contains 1.25 mg of conjugated estrogens. To make matters more worrying, they indicated that a soy burger has 1/10th the activity of a low-dose estrogen-containing oral contraceptive.

These results seem alarming, but I’m afraid the comparisons aren’t likely appropriate. First, the authors assume 10% bioavailability for the estrogenic components of the tofu burger. However, the 10% bioavailability is based on total isoflavone content and not the unconjugated portion (84-86).

This figure is highly variable between individuals, but just as importantly, the unconjugated portion of genistein is a small fraction of that 10% (varying from a fraction of a percent up to around 3%). Furthermore, considering the genistein content of that burger is around 20 mg, it makes it highly unlikely that the concentrations necessary for such strong estrogenic activity to occur in humans would occur after eating it.

Secondly, the authors’ method for calculating hypothetical blood values simply takes total blood volume and uses this to determine hypothetical blood values. However, this assumes that the soy isoflavones or E2Eqs remain in the blood and aren’t distributed anywhere else in the body. This is highly unlikely and is likely to substantially overestimate blood values.

While this may be a fun hypothetical exercise, the truth is that very large amounts are likely required for certain sensitive individuals to experience negative estrogenic effects from soy isoflavones in food.

Not only is the dose of soy isoflavones an important factor, but the duration of consumption may also be important, at least when it comes direct estrogenic effects.

For example, long-term soy consumption may increase production of equol (the more potent metabolite of daidzen), at least in some people (88). Additionally, in studies evaluating the potential estrogenic effects of soy isoflavones in women, an average of 13.4 weeks were required to achieve 50% of maximal effects and 48 weeks to achieve 80% (89).

Most people will want to know what the end results would be from consuming soy isoflavones. Would the decrease in testosterone be enough to impact their muscle mass? What about any direct estrogenic effects?

Unfortunately, there don’t seem to be any clear answers. The best I could do was find a study that supplemented untrained younger men undergoing resistance training with soy protein concentrate, whey protein concentrate, or placebo for 12 weeks (25).

This was a well-designed study. However, some potential weaknesses in this study include the lower dose of soy isoflavones (64 mg/day consisting of around 28 mg of daidzein and 36 mg of genistein as consumed in a soy protein concentrate consisting of around 80 g of protein), the untrained subjects, and the lack of pharmacokinetic data.

Additionally, while there was no statistically significant difference between the soy protein concentrate, whey concentrate, and placebo groups, only the soy group experienced a decrease in total testosterone. This small decrease was probably due to chance.

More likely, the doses administered were too low to have a negative impact on testosterone, but evaluating individual data might help identify if there’s potential support for the notion that there are more sensitive individuals that aren’t frequently represented in these studies (63).

The study also evaluated androgen receptor mRNA and AR protein in skeletal muscle, along with body composition and muscle fiber type cross-sectional area, but there was no difference between groups for lean mass gain or body composition.

In other studies, soy protein was shown to be inferior to whey protein. However, it’s believed that this is due to differences in protein quality, digestibility, and leucine content (93,94).

While body composition was unaffected by the protein supplements in the study above, whey protein concentrate had a large effect size upon type II muscle fiber cross-sectional area, while soy protein concentrate had a large effect size upon type I muscle fiber cross-sectional area (25).

In effect, whey protein increased type II muscle fibers (fast twitch) to a larger extent, while soy protein increased type I (slow twitch), an effect mirrored in animal models (95,96). This is an interesting finding itself, but it’s also supportive of the notion that the soy protein concentrate exerted an estrogenic effect because other studies showed a similar effect in women taking estrogenic oral contraceptives (97).

Some people who consume very high amounts of soy isoflavones can experience a negative effect on testosterone or androgen levels. These individuals are probably more prone to the effects of soy due to inter-individual differences in isoflavone metabolism and perhaps pharmacodynamic variables that make them more sensitive.

These folks are unlikely to experience negative effects of soy consumption except in cases where soy is a major source of their protein and they have high protein requirements. For example, a vegan bodybuilder who nearly exclusively relies on soy protein with significant amounts of soy isoflavones and consumes very large quantities.

These rare sensitive individuals might experience a sufficient estrogenic effect and a substantial decrease in T levels (a 50% decrease, although this also depends upon numerous variables, including baseline testosterone). As a result, you might expect a blunting of positive effects upon skeletal muscle from resistance exercise (101,102), but this is highly speculative.

To illustrate how rare this would be, you don’t have to look beyond studies comparing the effects of soy protein to whey or other forms. Soy protein generally performs adequately, and in cases where it does show inferior results, it’s likely due to differences in leucine content and protein quality.

Whenever there are discordant data in science, we must look for explanations. In this case, a reasonable hypothesis can help explain why studies evaluating the effects of soy protein on androgen levels in humans aren’t consistent:

• Dose-dependency. High amounts of soy isoflavones are likely required.
• Duration of exposure may be a factor.
• It may depend upon the individual, including unique genetic and environmental factors, along with unusually high consumption of soy isoflavones.
• If your protein supplementation consists entirely of a soy protein product with a significant amount of isoflavones, it could impact the hypertrophy experienced by muscle fiber type compared to other protein sources. This may be related to agonism at the ER-beta. Additionally, it may blunt the effects on androgen receptor expression from lifting compared to whey protein. Although, thus far, it doesn’t appear this negatively affects gains in lean mass. While some studies show differences in gains of lean mass when supplementing with whey versus soy, this is best explained by differences in protein quality and leucine content.
• Occasional consumption of soy protein or soy foods is unlikely to have a significant impact.
• Larger studies evaluating equol producer status, including pharmacokinetic data for genistein alongside testosterone levels and lean mass in trained subjects, would be useful to provide clearer answers.
• The greatest risk is likely in those who consume a high protein diet and get most of that from soy. Small effects on hormone levels in certain individuals likely require doses greater than 1.5 to 2 mg/kg of soy isoflavones and possibly much greater, including twice these values or more. This requires significant, practically unrealistic consumption of soy-based foods. Soy protein supplements also vary in their isoflavone content. See table below.
• Modern soy-containing beef alternatives like the Impossible Burger also are
  unlikely to contribute problematic amounts of soy isoflavones.
• Effects on hormone levels might include a decrease in total testosterone and free testosterone. But these effects would be small and unlikely to be clinically meaningful. Very large doses (probably in the range of 3-8 mg/kg or higher of genistein, for example), and in only certain individuals, may have significant enough effects to impact well-being. When factoring in protein-binding, doses may need to be in the range of 6 mg/kg or higher for genistein.

Bottom line, unless you’re a rare individual, eating the occasional soy-heavy meal isn’t going to affect you to any noticeable degree. But avoid soy protein isolates, just as much for their comparatively poor leucine content as their relatively high concentration of soy isoflavones.

Nevertheless, don’t let anyone tell you that the notion that soy can affect testosterone and estrogen is a myth.

* Milligrams per 100 gram serving.